By Arnout Schutte, Co-founder / Managing Director
Agentic commerce starts with product data. Not with AI.
AI agents are taking over more steps in the customer journey: searching, comparing, recommending and, increasingly, purchasing on behalf of consumers.
But one thing does not change: an AI agent can only work with product information it can find, understand and trust. If information is missing, contradictory or impossible to verify, a product becomes harder to compare and may ultimately be left out of the recommendation.
That is why the ShoppingTomorrow PIM & AI expert group explored how organisations can prepare their product data for a world in which AI is becoming an increasingly important part of commerce. ConnectingTheDots and Truvio host the expert group, with Squadra acting as chair.
Data first. Then AI.
When organisations start exploring AI, they often focus first on the visible applications: generating product descriptions, creating images, translating content or building an AI assistant.
The most important conclusion from the bluepaper comes one step earlier:
Start with the data and processes AI needs to trust.
Product information needs to be complete, accurate, current, consistent and traceable. Only then can AI reliably support the collection, validation, enrichment and distribution of product data.
That is where the real difference is made. Not by adding another standalone AI tool, but by improving the product data flow underneath it.
From supplier to AI agent.
The bluepaper follows product data across the entire flow. From collecting and validating supplier data to enriching and publishing it to webshops, marketplaces and emerging AI channels.
- Import & Onboarding: use AI to recognise, structure and validate incoming supplier data.
- Data quality: identify missing information and inconsistencies earlier in the process.
- Enrichment: enrich attributes, context, content and media at scale.
- GEO: structure product information so AI systems can interpret and substantiate it more effectively.
- Compliance: bring sustainability, provenance and compliance data into the core product data model.
- Agentic commerce: prepare product information for AI assistants and agents that search, compare and recommend products.
20 million records a day. Live within ten minutes.
The bluepaper combines these principles with real-world cases from organisations including Intergamma, The Sting, DMG, Sligro, Wiltec and Azerty.
Azerty processes around 20 million records every day from approximately 100 data sources. With ConnectingTheDots, product information for more than 400,000 products is processed centrally and automatically distributed to webshops, ERP, its data lake and marketplaces.
A new product can be live with the correct specifications within ten minutes.
Adding AI is not the goal. Making product data flow is.
If an AI agent cannot see it, it does not exist.
The shift towards agentic commerce makes high-quality product data even more important.
Product information will no longer need to be understandable only to people. It also needs to be understandable to machines that independently select, compare and recommend products.
The best-written marketing copy will not automatically win. Structured, consistent and traceable product information gives AI systems far more context and confidence.
That changes the role of PIM. It becomes more than a place to store product information. It becomes the foundation AI can work from.
Good product data starts before your PIM. And good AI starts with good product data.
Ready for agentic commerce?
Discover what AI requires from your product data, PIM and organisation.
Download the bluepaper or discover ConnectingTheDots.
This bluepaper was developed by the ShoppingTomorrow PIM & AI expert group, with ShoppingTomorrow, ConnectingTheDots, Truvio and Squadra. On behalf of ConnectingTheDots: Arnout Schutte (Managing Director & Founder) and Max Schrevelius (Commercial Director & Founder).